Two years ago, I wrote this post on the possible areas that I see for ethereum + AI intersections: https://t.co/ds9mLnrJWm This is a topic that many people are excited about, but where I always worry that we think about the two from completely separate philosophical

Ring paid somewhere between $8 and $10 million for a 30-second Super Bowl spot to tell 120 million viewers that their cameras now scan neighborhoods using AI. The math is wild. Ring has roughly 20 million devices in American homes. Search Party is enabled by default. The opt-out

Great read! Berk has helped me so much. If you’re raising reach out to him

AI in non dev-tools are surprisingly volatile. @figma was great, @framer and @canva not so much. I’m excited to try more tools next week

Before our launch, I really struggled with narrowing down our ICP because I could see use cases for data science teams, software engineering teams, and PMs/analysts. I’ve had way more customer conversations post-launch, and it’s made narrowing down ICP much easier

Building an autonomous MLE is fun because models are already non-deterministic, and using non-deterministic software (i.e. LLMs) to make better non-deterministic software (i.e. models) is fun

We launched last week and there were way more bugs that I would have imagined. Edge cases that hadn’t been exposed, customers tried things that we didn’t think they would try, and our agent responding in ways that we hadn’t seen before

I’m starting to keep track of important things I learned every week. Naturally, I asked my conversation buddies (aka Gemini & Claude), and got back stupids things like “You stopped trying to convince people your AI was good with adjectives” 😂 Now onto some real learnings 🧵

I know the PFHOF was upset about the leaks, but it was better that Belichick missing out leaked then, so it did not completely overshadow those who were announced as elected tonight. They deserve to be celebrated.

To test it, we entered a Kaggle competition. 31,791 participants. Our autonomous agent placed rank 782. Top 2.5%. So yeah, pivoting sucks. But it’s also the best decision we made and consistently talk about how it was the right decision 😀

So we built an autonomous ML engineer. - Describe your problem in plain English - Connect your data - Agent builds, trains, and deploys your model No ML expertise required

Then we found the real bottleneck: Not enough MLE talent - Engineering teams drowning in model requests - PMs/analysts blocked from building models - Scientists at companies with zero MLEs The problem wasn't compute. It was a lack of talent

The other 10% didn't care that our compute was super cheap. Price alone doesn't win customers. Switching costs, preferences, product requirements — these all matter more than being "much cheaper." We weren't better. We were just less expensive.

For a while, I felt like I had personally failed. We had built something. Customers were using it. But deep down, I knew it wasn't working. We hit a wall. 90% of customers just wanted to use Claude, Gemini, or GPT. They didn't care about open source hosting.

As a founder, you want your first idea to be the billion-dollar one. You want to be the next Google or Facebook immediately. But that’s not reality. Most great companies pivot. We just did. It was painful, but it saved us. Here is the story 🧵👇

As a founder, I constantly think about product-market fit. How will I know when I have it? What will it feel like? And it'll probably feel like this. I just bought a razor from @HensonShaving. I love it! I use to use @Gillette Mach3s, but henson is way way better. No bumps or

Have been following reactions to what I said about L2s about 1.5 days ago. One important thing that I believe is: "make yet another EVM chain and add an optimistic bridge to Ethereum with a 1 week delay" is to infra what forking Compound is to governance - something we've done

warriors of the east… without the talent 😂

LOL

One fun part of being a founder is that the responsibilities are so broad I end up working on a bunch of different things like engineering, design, etc. So I spent the last few days trying to put together a one-pager for potential customers. I tried a few different products,